Messages in this thread | | | Date | Sun, 7 Dec 2008 23:09:58 +0530 | From | Balbir Singh <> | Subject | Re: [PATCH] introduce get_mm_hiwater_xxx(), fix taskstats->hiwater_xxx accounting |
| |
* Oleg Nesterov <oleg@redhat.com> [2008-12-07 18:28:45]:
[snip]
> No. When the task does unmap, vm does update_hiwater_vm() "internally", > it does not need the help from do_exit(). And do_exit() can't help, > it is to late to calculate the maximum, ->total_vm was already decreased. > > do_exit() itself does not affect rss/vm. Until we call exit_mmap(), > but at this point ->mm is dead, nobody can look at it, its ->mm_users > is zero. >
OK, I think that makes sense, since free would have cleaned up the memory before entering the exit system call.
[snip]
-- Balbir
| |